Relationships between hepatic neoplasms and related lesions and exposure to toxic chemicals in marine fish from the U.S. West Coast.

English sole (Parophrys vetulus) inhabiting polluted waterways and embayments of Puget Sound, Washington, are affected with a variety of multiple, co-occurring idiopathic hepatic lesions, including unique degenerative conditions, putatively preneoplastic foci of cellular alteration, and neoplasms. Results of a statistical analysis of the patterns of co-occurrence of these lesions in wild English sole are consistent with the concept that these lesions represent morphologically identifiable steps forming a sequence of progression ultimately leading to the development of hepatic neoplasms. This progressive sequence parallels the pattern identified in experimental models of chemically induced hepatocarcinogenesis in rodents. The rationale for the hypothesis that these lesions in wild English sole can be caused by exposure to certain hepatoxic and hepatocarcinogenic xenobiotic compounds in the marine environment is based on the demonstration of significant and consistent statistical associations between levels of aromatic hydrocarbons (AHs) in sediment and prevalences of these idiopathic liver lesions; a significant contribution by sediment AHs to the variability in hepatic neoplasm prevalence in a logistic regression model; significantly increased probabilities for several idiopathic lesions in sole from chemically contaminated sites in Puget Sound; significant correlations between uptake of polycyclic aromatic hydrocarbons, as measured by levels of fluorescent metabolites of aromatic compounds in bile of sole, and prevalences of several hepatic lesion types; and experimental induction of unique degenerative, proliferative, and putatively preneoplastic focal lesions in English sole injected with either benzo(a)pyrene or a polycyclic aromatic hydrocarbons (PAH) enriched fraction of an extract from a contaminated urban sediment from Puget Sound.(ABSTRACT TRUNCATED AT 250 WORDS)

steps forming a sequence of progressn ultimately leading to the development of bepatic neoplas This progressive sequence parailels the pattern identified in experimental models of cbemically induced hepatoarcinogenesis in rodents. The rationale for the hypothesis that these lesins in wild Engis sole can be caused by exposure to certain hepatoxic and hepatocarcinogenic xenobiotic compounds in the marine environment is based on the demonstration of ign nt and consistent s l associations between lvels of aromatic hydrocarbons (AHs) in sediment and prvalences of these idiopathic ver lesions; a significant contribution by sediment AHs to the variability in hepatic neoplasm prevalence in a gistic model; ntly increed probabilities for several idiopathic lesions in sole fromcbemicaly contaminated sites in Puget Sound; ignifiant rdations between uptake of polydic aro tic h d as measured by levels offluorescent metabolites of aromatic compounds in bile of sole, and prevalences ofseeral hepatic lesion types; and experimental induction of unique degenerative, prolferative, and putatively preneoplastic focal lesions in English sole inected with elther benzo(a)pyrene or a polycyclic aromatic hydrocarbons (PAH) enriched fraction ofan extract from acontmited urban sediment from Puget Sound. Simlar pes of idiopathic hepatic lesions have been detected in white Introduction For the past 15 years, our laboratory has been engaged in fish pathology research in Puget Sound and other marine coastal areas ofthe United States, using a multidisciplinary approach to study the deleterious effects in fish resulting from exposure to sediment-associated xenobiotic chemicals. This strategy combines histopathologic examination ofresident fish with chemical analyses of sediment and various fish tissues and bile and analyses of several activation/detoxication enzymes in fish tissues. This sampling design was developed subsequent to the exposure to sediment-associated chemical contaminants. To date, of the identifiable sediment-associated chemicals, polycyclic aromatic hydrocarbons (PAHs) appear to be the most likely chemical etiologic agents for these liver lesions.
The vertebrate liver is well recognized as a target organ for effects of xenobiotic chemicals and carcinogens; we have therefore directed our efforts in the area of fish pathology to detecting potentially toxicopathic liver lesions in wild fish, especially English sole. Because certain hepatic lesions in wild fish resemble conditions induced in rodents and fish by controlled exposure to toxicants and carcinogens, these lesions represent potential indicators of effects from exposure to xenobiotic chemicals in the environment. The advantage of this approach, when combined with supportive data demonstrating exposure to and uptake of xenobiotic chemicals, is that the presence of liver lesions in wild fish that are morphologically identical to those induced by hepatotoxins/hepatocarcinogens in lab studies provides a direct method of assessing the potential sublethal effects of xenobiotics on wild fish. The major goal of our research has been to investigate the etiology of liver neoplasms and other related lesions, primarily in bottomfish species in contact with contaminated sediments, and by doing so, determine the significance of these lesion types as reliable biological indicators of effects of toxic/carcinogenic chemical exposure. To date our research goals have not specifically included the investigation of human health effects resulting from consumption ofchemically contaminated fish. However, the results ofour studies have a strong bearing on studies related to human health risk because they demonstrate the exposure of commercially and recreationally important fish species to chemical toxicants and carcinogens.

Hepatic Pathology of Wild English Sole
Ofthe species studied in Puget Sound, English sole is the most useful indicator species and displays a spectrum ofmultiple, cooccurring idiopathic liver lesions ( Table 1) that is quite similar to the variety of hepatic lesions induced experimentally by chemical hepatotoxins and carcinogens in the mouse, rat, and certain fish species (11). These lesions are found almost strictly, or at significantly higher prevalences, in English sole captured from highly polluted urban estuaries and embayments in Puget Sound as compared to relatively unpolluted reference stations (2,(5)(6)(7)(8)(9)(10). This geographic pattern of lesion distribution in English sole is closely parlleled by other bottom-dwelling species in Puget Sound, including rock sole (Lepidopsetta bilineata) and starry flounder (Platichthys stellatus), but lesions are observed at lower prevalences in these species (12). By virtue of their morphologic similarity to lesions induced by experimental exposure to hepatotoxins and hepatocarcinogens in rodents and fish and the geographic distribution ofaffected fish within Puget Sound, we have hypothesized these lesion types to be a likely result of contaminant exposure.
The pathologic anatomy of these lesion types in English sole has recently been comprehensively described (.,). The most frequently encountered idiopathic hepatic lesion, appearing first in young of the year or juvenile fish (10) is a unique, nonproliferative degenerative condition characterized by a marked increase in nuclear and cellular diameters in affected hepatocytes, nuclear hyperchromasia, and various cytoplasmic degenerative changes including hyalinization and hydropy. Because there is no appreciable accompanying cellular inflammatory response, this lesion is termed megalocytic hepatosis. Megalocytic hepatosis and its related precursor condition, nuclear pleomorphism, are identical to the early changes induced by a spectrum of hepatotoxicants and carcinogens in rodent models and are interpreted as manifestations of the chronic to subchronic hepatotoxicity of these compounds. Such hepatic degeneration and necrosis is known to be an essential early step in hepatic carcinogenesis, and functions to provide a stimulus for the compensatory proliferative response that actually fixes the molecular lesions in the hepatocellular DNA that are responsible for neoplastic transformation and initiation (13).
The lesion that most frequently co-occurs (11) with megalocytic hepatosis is hepatocellular regeneration. This lesion represents the compensatory proliferative response to the degeneration and necrosis manifest in megalocytic hepatosis that is a necessary step in the process of hepatic neoplasia (14,15). These regenerative cells may be the population ofproliferative cells from which initiated hepatocytes develop, according to the two-stage initiation-promotion model of hepatocarcinogenesis, and may represent a population of cells that are resistant to the cytotoxic effects (16) of the carcinogens present in the polluted sediments in which the English sole resides. Also frequently co-ccurring with megalocytic hepatosis and regenerative hepatocytes are discrete, spherical foci or micronodules ofthree basic types (11), based on their tinctorial qualities in the standard hematoxylin and eosin stain. These foci ofcellular alteration generally first appear in fish at least 1 year ofage (10). These foci are considered to be preneoplastic lesions in rodent hepatocarcinogenesis models (17), from which true neoplasms may develop under the correct conditions and which represent an obligatory precursor step in the induction ofhepatic neoplasms (18). The borders ofthese focal lesions blend into the surrounding parenchyma, and there is no compression. The three basic types offoci ofcellular alteration seen in English sole are the eosinophilic, basophilic, and clear cell focus. All three types of foci can be recognized histochemically by a reduction or absence in cytoplasmic iron in both English sole (11) and rodents (19) and rarely contain other hepatic elements such as bile ducts, pancreatic acini, blood vessels, or melanomacrophage centers (macrophage aggregates). The relationship among these foci types in terms of their role and fate in hepatocarcinogenesis is not well understood. However, it is commonly believed that the basophilic focus is the focal lesion most proximate, in the temporal sense, to the development of true hepatocellular neoplasms in rodents (20,21) and is considered a small hepatocellular carcinoma in the rainbow trout model (22).
Substantial prevalences (up to 25 %) ofhepatic neoplasms are also found in adult English sole generally at least 2 years of age (10) and almost strictly from polluted sites in Puget Sound. These lesions are typically obvious grossly (11), and often multiple nodules form the bulk ofthe liver. The basic histologic types of hepatic neoplasms found are, of necessity, classified strictly on their morphologic appearance because we have little information on their biological behavior and no information as to the eventual clinical outcome in affected fish. Liver cell adenomas are characterized by compression of the surrounding parenchyma, well defined separation ofproliferative tissue from normal tissue, fairly normal architecture, and relative absence ofother hepatic elements. Liver cell adenomas are usually basophilic, but eosinophilic and vacuolated variants do occur. As is the case for all hepatocellular neoplasms in English sole, the composite hepatocytes are resistant to iron accumulation (11).
Hepatocellular neoplasms with accepted histologic features of malignancy are also found in English sole. These hepatocellular carcinomas are typically multiple, with irregular borders, and show compression and invasion. They have an irregular trabecular architecture and are composed ofpolygonal, pleomorphic, anaplastic, and occasionally vacuolated hepatocytes showing loss of cellular polarity with respect to their orientation toward the sinusoids. Occasionally found within adenomas are areas of hepatocytes with a morphology consistent with a carcinoma. These nodules within nodules suggest progression from an adenoma to a carcinoma as one possible pathway ofcarcinoma development (11).
A distinct class of hepatic neoplasms originate from epithelial cells of the intrahepatic biliary system. These neoplasms are characterized by a proliferation of biliary epithelium assuming relative degrees of tubular organization. Well-organized and benign-appearing tumors of biliary origin are rare, and the cholangiocellular carcinoma composed ofirregular tubular profiles that invade and insinuate into the surrounding hepatocellular parenchyma is typically found. All neoplasms ofthis type are invasive, replace high proportions of the parenchyma, and can metastasize (11). We can therefore confidently diagnose tumors with similar morphology as carcinomas. The histologic and ultrastructural detail ofthis class ofneoplasms in English sole has recently been described (23). Occasionally we detect carcinomas with a mixed biliary and hepatocellular origin (11).
No evidence of a viral etiology for any of these lesions in English sole has been revealed through ultrastructural observation by transmission electron microscopy (23,24).

Patterns of Hepatic Lesion Co-Occurrence in Wild English Sole
Livers of English sole from polluted sites in Puget Sound are affected by a spectrum ofmultiple, co-occurring idiopathic liver lesions that closely parallels (11) the degenerative, regenerative, preneoplastic, and neoplastic hepatic lesions induced experi-mentally by chemical hepatotoxins and hepatocarcinogens in the mouse, rat (7) and certain fish species (22,25). To clarify the role of these lesion types within the context of the process of liver neoplasia in English sole, we conducted a statistical analysis of lesion type co-occurrence in English sole from Eagle Harbor, Washington (11), a site with sediments severely polluted by polycyclic aromatic hydrocarbons (4) released from a nearby creosote plant (26). The inferences from this analysis are that lesions which co-occur at levels of statistical significance may be caused by etiologic agents that are associated with each other and/or may be temporally related to one another. One possible temporal association would include involvement in a series oflesions progressing toward hepatic neoplasms, as has been clearly demonstrated in rodent hepatocarcinogenesis models (17,27). This type ofanalysis represents an attempt to interpret, in a temporal sense, the profusion of multiple hepatic lesion types typically encountered in wild English sole, a pattern consistent with what one would expect under a situation of continuous exposure to chemical hepatocarcinogens (28).
The significant patterns emerging from this analysis are shown in Figure 1 and are consistent with the experimentally determined temporal histogenesis ofchemically induced hepatic neoplasia in the mouse and rat (17) and rainbow trout (22). Megalocytic hepatosis is viewed as the initial degenerative lesion resulting from the cytoxic effects ofthe hepatocarcinogens English sole are exposed to in the sediment. This is because megalocytic hepatosis appears first in fish less than 1 year of age (10), and is the most common hepatic lesion detecteJ in juvenile sole (10; Myers, unpublished observations), is associated with other degenerative and necrotic conditions of the liver (11), is the most common idiopathic lesion detected injuvenile or adult sole from contaminated sites (10,11; Myers, unpublished observations), and appears identical to the initial lesions resulting from experimental exposure to certain hepatocarcinogens in the mouse, rat, and certain fish species.
Megalocytic hepatosis is strongly associated with hepatocellular regeneration, manifesting the compensatory proliferative response to the liver degeneration and necrosis seen in and associated with this lesion, and which is an essential step in neoplasia in chemically induced, experimental hepatocarcinogenesis in rodents (15). Megalocytic hepatosis is also associated with all three major types of foci ofcellular alteration (FCA) (clear cell, eosinophilic, and basophilic focus), which are considered preneoplastic lesions in rodent hepatocarcinogenesis models from which true neoplasms may develop under the correct conditions (18). In rodent hepatocarcinogenesis models, development of these focal lesions represents an obligatory precursor step in the induction of hepatocellular neoplasms. Focal lesions occur earliest in sole at least 1 year of age and are found at higher prevalences in older adults (10). Based on these strong associations, the age ofaffected fish, and the established temporal histogenesis of liver neoplasia in rodents, we believe that the regenerative and preneoplastic focal lesions follow the initial hepatotoxic lesion, megalocytic hepatosis.
Strong, consistent associations also exist among all of the preneoplastic FCA and between these focal lesions and hepatic neoplasms, especially the liver cell adenoma. These patterns strongly suggest that the focal lesions in English sole are, in fact, preneoplastic lesions. The lack ofa strong association between megalocytic hepatosis and the hepatic neoplasms also suggests that these lesions are sufficiently separated in the temporal sequence of lesion progression that they tend not to co-occur. In summary, this analysis strongly suggests that the cooccurring lesion types observed in English sole from polluted areas of Puget Sound comprise a temporal sequence of lesions progressing toward hepatic neoplasms that parallels the same process in experimental mouse and rat hepatocarcinogenesis. These results are the first confirmation, in any wild vertebrate population exposed to hepatotoxins and carcinogens in the environment, of the experimentally derived histogenesis of chemically induced hepatic neoplasia (11). This evidence also indirectly suggests that xenobiotic chemical hepatotoxicants and carcinogens in the environment may be the inducers ofthese lesions in English sole liver.
This model also strongly indicates the need to consider hepatic lesion types other than neoplasms as indicators ofeffects ofcontaminant exposure when conducting fish liver histopathology monitoring studies, especially when individuals of the target species are young juveniles (0+ or 1+, age in years). Results from recently completed field studies on bioindicators of contminant exposure injuveniles ofthree flatfish species (English sole, rock sole, and starry flounder) from Puget Sound further indicate that the hepatic lesion types megalocytic hepatosis/nuclear pleomorphism and foci ofcellular alteration are reliable indicators ofexposure to hepatotoxic and hepatocarcinogenic compounds, especially in the two former species (Myers, unpublished observations).

Statistical Associations among Liver Lesions and Indices of Contaminant Exposure
The results thus far presentd support the hypothesis that liver lesions are the result ofexposure to environmental contminants by virtue ofthe similarity of these lesions to lesions induced by experimental exposure to hepatotoxicants and carcinogens/and by the unique geographic distribution ofthese conditions among the sampling sites in Puget Sound. Idiopathic hepatic lesions are detected primarily in fish from estuaries and embayments in close proximity to urban centers or major sources of chemical contamination (2)(3)(4)(5)(6)(7)(8)(9)(10)29). In a more direct approach to this issue we have documented several strong associations between these lesions and the presence ofand exposure to certain chemical contaminants in the environment.
PRsitive statistical associations ofthese lesions with contaminant exposure have been demonstrated by comparing lesion prevalences with actual concentrations ofaromatic hydrocarbon contaminants in sediment by the Spearman's rank correlation method. Using this approach, we have shown significant correlations between overall idiopathic liver lesion prevalence in fish and concentrations ofsummed aromatic hydrocarbons (AHs) in sediment (2). This same association has subsequently been established between hepatic neoplasms and summed AHs (2,6); foci of cellular alteration and summed AHs (30) ( Table 2); and megalocytic hepatosis and summed AHs (30) ( Table 2).
A fundamental tenet of analytical epidemiology maintains that, while the strength of a particular statistical association in one study is important in establishing a relationship, ofperhaps greater importance is the consistency ofthat relationship among multiple studies of a similar type. Accordingly, we have demonstrated a consistently positive association between liver neoplasm prevalence and sediment-associated AHs across six separate field studies (30). A strong, consistent association ofthe same type also exists for megalocytic hepatosis and for foci of cellular alteration (Table 2).
We have also developed a logistic regression model (Fig. 2) that incorporates neoplasm prevalence, sediment conmiinant levels, interactions among different classes ofchemicals, and fish size (an indirect measure of fish age) from multiple studies in Puget Sound (30,31). These results further strengthen the hypothesized role ofAHs in sediment as a major etiologic agent of liver neoplasms in English sole, as the variation in levels of chemicals in this group accounts for about 12% ofthe variation in prevalence ofhepatic neoplasms, more than for any other identifiable conuminant group. The model also points out the significant positive influence offish size (age), and to a lesser extent, PCB contmination on neoplasm prevalence. However, the fact that less than 35 % of the variation in neoplasm prevalence can be explained by this model suggests the importance of other potential factors such as fish movement, unidentified chemicals, and metabolites that may be involved in the explanation of the variability in prevalence ofthese neoplasms. For example, it is estimated that we can currently and reliably identify only a fraction of the chemical contaminants present in sediments (U. Varanasi, personal communication). We have also documented significant correlations between levels of fluorescent aromatic compounds (FACs) in bile of English sole [including metabolites of BaP, pyrene, fluorene, fluoranthene, dibenzofuran, and phenanthrene (32)] and prevalences ofliver lesions in sole from multiple sampling sites (29,33). Although the positive correlation between these bile metabolites and neoplasms has received much attention, equally in FAC levels result from alterations in PAH metabolism, bile production, or other biliary function caused by the presence of hepatic lesions. For example, within a site, bile metabolite levels in sole with or without lesions are generally not significantly different, regardless of lesion type considered (30). strong correlations exist between these metabolites and prevalences of megalocytic hepatosis and foci of cellular alteration (Table 3). This is apparently due to the fact that the associations shown among the lesion types and metabolite levels are not truly direct. Bile metabolite levels simply provide a more direct measure than sediment contminants ofrecent exposure to PAHs, and are not a direct reflection or predictor ofthe presence of lesions. Furthermore, no evidence exists to suggest that differences  Consequently, bile FACs levels are interpreted as being linked to lesion prevalences only by virtue oftheir nature as indicators of recent contaminant exposure, uptake, and metabolism; they do not reliably predict the presence of specific hepatic lesions in individual fish. Additional support for the hypothesis ofa chemical contaminant etiology for hepatic lesions in wild English sole is derived from an epidemiologic analysis ofa multiyear study (1979)(1980)(1981)(1982)(1983)(1984) investigating the significance ofpotential risk factors associated with lesion prevalences in English sole, including age, gender, and site, season, and year ofcapture (10). Table 4 summarizes the important pertinent results ofthis study, showing odds ratios for particular risk factors as applied to the major lesion categories. Certain risk factors inherent to the Duwamish Waterway (e.g., xenobiotic contaminants) significantly influence the prevalence of not only neoplasms, but also ofpreneoplastic foci of cellular alteration, and the related degenerative conditions of nuclear pleomorphism and megalocytic hepatosis. Other less contaminated urban sites (2) in Elliott Bay (N. Seattle waterfront, S. Seattle waterfront) show elevated odds ratios only for nuclear pleomorphism and megalocytic hepatosis. Fish age is also shown as a significant risk factor for all lesion categories with the exception ofthe latter two lesion types. In summary, the results of this epizootiologic study demonstrate the lack of influence offish gender on the probability of hepatic lesion development; the relative insignificance of season or year of capture as significant factors for most of the lesion types; the positive influence offish age and implied increased time of exposure to chemical contaminants on the probability of developing preneoplastic and neoplastic hepatic lesions; the earliest age at which sole become affected with specific lesions; and the significandy increased risk of being affected with certain lesions for sole from chemically contaminated areas of Puget Sound.

Chemical Induction of Idiopathic Hepatic Lesions in English Sole
The types of evidence discussed earlier supporting a xenobiotic chemical etiology for neoplasms and other associated liver lesions in English sole rely mainly on demonstrating statistical correlations and associations and are not generally interpreted as proof of a direct cause-and-effect relationship. We have recently addressed this question directly by conducting a series of long-term laboratory exposure studies with English sole, in which sole were injected with a PAH-enriched fraction of an extract from a contaminated sediment, or a model hepatocarcinogen (BaP), with appropriate controls (31,34). To briefly summaize this work, only injection of an extract from Eagle Harbor and the model hepatocarcinogen, BaP, induced high incidences of the unique hepatotoxic lesions, nuclear pleomorphism (up to 83%) and megalocytic hepatosis (up to 60%), and associated degenerative lesions. The most important result ofthis study was the induction, again only in the exposure groups mentioned above, ofbasophilic foci ofcellular alteration. This lesion type is an essentially proven preneoplastic lesion in rodent hepatocarcinogenesis models, is thought to be the immediate precursor ofhepatic neoplasms in these models, and is considered a microcarcinoma in the model for rainbow trout. All of these lesion types were induced within 18 months after the initial exposure. The results from this laboratory exposure study directly implicate certain chemical contaminants in the sediments, especially PAHs, as the agents of hepatocellular nuclear pleomorphism/megalocytic hepatosis, regeneration, preneoplastic foci of cellular alteration, and, at least by association, hepatic neoplasms in English sole from Puget Sound.
Levels of PAHs in Liver Tissue of English Sole from Puget Sound A comprehensive study specifically investigating the relationship between the presence or absence ofspecific liver lesions and levels ofPAHs in liver tissue ofthe same English sole has yet to be conducted. The premise for this type of study rests on the hypothesis that lesion presence is a predictor of tissue burdens ofPAHs. The primary reason such studies have not been done is due to the well-known fact that fish effectively metabolize parent PAHs (35) to compounds that are not detectable by routine methods. For this reason, metabolites ofPAHs measured as FACs are routinely assayed in the bile of English sole, and the relationship between prevalences of hepatic lesions and FACs levels has been discussed previously. PAH levels in liver tissue that have been reported are typically very low, quite variable within a species at a particular site, and no clear relationship between the presence ofliver lesions and levels ofPAHs in liver tissue from the same fish has been established. For example, earlier studies in Puget Sound (2) showed that the concentration of individual AHs in liver tissue of English sole was usually below or close to the limit of detection for samples from either the polluted sites of Hylebos Waterway (Commencement Bay) and the Duwamish Waterway (Elliott Bay), or the nonurban reference sites of Case Inlet and Port Madison. However, in a limited number ofEnglish and rock sole liver samples analyzed, PAHs were detected in some samples, with the most commonly detected compounds being benzothiophene, dibenzothiophene, pyrene, and chrysene (8). However, because of the extensive metabolism ofPAHs in the teleost liver (35), the level(s) ofparent compounds in this organ is not considered a reliable measure of PAH exposure. The quantitation of levels of genotoxic compounds (including meabolites ofPAHs) covalendy bound as adducts to hepatic DNA of English sole and other species (36) may eventually prove to be a reliable hepatic indicator of exposure to genotoxic PAHs.

Levels of PAHs in Muscle of English Sole from Puget Sound
Although the results of our field and laboratory research strongly indicate that sediment-associated PAHs can be an etiologic agent ofthe hepatic lesions seen in wild English sole, a direct extrapolation of these findings to a presumed human health risk from consumption of muscle from these same fish cannot yet be made, primarily because ofthe paucity ofreliable data on levels ofPAHs and/or their metabolites in edible flesh of English sole (37)(38)(39)(40).  (41) that are similar to those seen in English sole from Puget Sound. However, the prevalences of these conditions are much lower in white croaker. Table 5 summarizes these data for some of the more important lesion types, showing a high prevalence of megalocytic hepatosis/nuclear pleomorphism (MH/NP) at Cerritos Channel and Reservation Point and lower prevalences of either the putatively preneoplastic focal lesions [eosinophilic foci/basophilic foci (EF/BF)] or hepatic neoplasms at Cerritos Channel, Queensway Bay, Reservation Point, and the Hyperion sewage outfall. Most of the detected neoplasms were tumors of the intrahepatic biliary elements. Although the evidence linking these lesions to exposure to chemical contaminants is far less conclusive than in the situation for English sole from Puget Sound, the lesions in white croaker were generally found at sites with more contaminated sediments, especially those with high levels of PAHs (Table 6). However, lesion prevalences among the sites, especially for the neoplasms and focal lesions, were low and generally not statistically distinguishable (Table 5). In general, the sites showing highest levels ofBaP-like metabolites in croaker bile (Table 7) were also sites exhibiting non-zero prevalences ofhepatic focal lesions, neoplasms, or megalocytic hepatosis ( Table 5).
The etiology of liver lesions in white croaker from this area remains an open issue, and no strong associations between exposure to aromatic hydrocarbons and the presence of liver disease have been established to date. In response to the potential problem regarding human exposure to elevated chlorinated hydrocarbons in the flesh of white croaker and other sportfish from this area, a number of studies have addressed the issue of contaminant levels in edible flesh ofwhite croaker (42)(43)(44). No study has yet been conducted specifically examining the relationship(s) between idiopathic hepatic lesion presence/absence and tissue burdens of chemical contaminants in white croaker. Recently, the state of California has been funding research on these potential health impacts ofconsumption ofchemically contaminated fish (including white croaker) by sportfishermen. To our knowledge, this study has not been completed and no results have yet been released.

Summary
Results from field and laboratory studies investigating the etiology of hepatic neoplasms and other lesions involved in hepatic neoplasia in English sole support a causal relationship between exposure to sediment-associated conaminants, primari-. ly polycyclic aromatic hydrocarbons, and the development of these hepatic lesions. Major lines of evidence supporting this view are: a) construction ofa model ofthe histogenesis ofhepatic neoplasia in English sole that parallels the multistep neoplastic process demonstrated in experimental hepatocarcinogenesis models; b) the development of a logistic regression model that accounts for about 35 % ofthe variation in neoplasm prevalence in wild English sole and attributes 12 % of the total variation to levels ofPAHs in sediment; c) the creation ofa separate logistic regression model showing significant increases in the probability ofmost hepatic lesions occurrng in sole from contaminated sites and in older fish exposed to contaminants for longer periods; d) the demonstration of significant positive correlations between FACs levels in bile and prevalences ofhepatic lesion types in wild English sole; and e) the experimental induction of unique hepatotoxic, regenerative, and putatively preneoplastic focal lesions in English sole by repeated and extended exposure to a model hepatotoxicant/hepatocarcinogen (BaP) or a PAHenriched extract ofa contaminated sediment from a site in Puget Sound.
Research on hepatic neoplasia in English sole has been greatly facilitated by the existence ofa kind ofnatural laboratory in Puget Sound from which has been developed a strong epizootiological database that implicates xenobiotic contaminants in the etiology ofneoplasms and other related hepatic lesions. This information has led us to certain hypotheses that have been tested and at least partially verified in experimental laboratory exposure studies. Consequently, our studies on hepatic neoplasia in English sole are not affected by the central problem inherent to other fish and rodent hepatocarcinogenesis models; namely, the inability to relate results of laboratory studies directly to those derived from observations in the environment.
However, regardless of the significant progress thus far achieved in elucidating aspects of the etiology of hepatic neoplasia in English sole, more research is certainly needed on the multiple genetic and epigenetic factors and the mechanisms that are operant in the process ofhepatic neoplasia in this species. The fact that we can currently identify factors that account for only 35 % ofthe variation in neoplasm prevalence in wild English sole in the logistic regression model points to the obvious need for further investigation into etiological factors not explicitly included in this model. These factors could include xenobiotic chemical agents not currently identifiable that may act as carcinogens, co-carcinogens or promoters, the role of these or other factors in activation of protooncogenes, etc. Greater knowledge of these additional influencing factors would contribute to a strengthening of the model we have partially developed, and would help guide further laboratory studies delineating the mechanisms comprising the multistep pathway ofhepatocarcinogenesis in English sole that can be extrapolated to other fish species.
The issue of whether or not a human health risk exists due to consumption of English sole and other fish species affected by hepatic neoplasms and/or related lesions cannot be resolved from the types and amounts ofevidence currently available. Although reported levels in muscle tissue of the xenobiotic chemicals (PAHs) shown to be important etiologic agents of hepatic neoplasia are very low or absent in this species, such data are fragmentary and often unreliable. Moreover, fish extensively metabolize PAHs (35); consequently, it must be a future goal to develop more sensitive and reliable methods for detection of these metabolites in fish muscle.